The problem with metrics is a big problem for AI
This article talked about some problems in AI and put forward some ways to improve it. We can’t measure the things that matter most. Metrics can, and will be gamed. Metrics tend to overemphasize short-term concerns. Many metrics gather data of what we do in highly addictive environments.
One way yo keep metrics in their place is to consider a slate of many metrics for fuller picture. Data should be combined with listening to first-person experience of those working at these companies. Another key to keeping metrics in their proper place is to keep domain experts and those who will be most impacted closely involved in their development and use.
Inadvertent Algorithmic Cruelty
The writer used his own experience to illustrate the phenomenon of inadvertent algorithmic cruelty and possible ways to avoid it. He emphasized the importance of empathetic design. He promote people to increases awareness of and consideration for the failure modes, the edge cases, the worst-case scenarios, which are definitely important aspect for designers to consider and think about.